Bug 783982

Summary:	[3.3.0qa19] Possible fd leak when multiple graphs are opened
Product:	[Community] GlusterFS	Reporter:	Anush Shetty <ashetty>
Component:	core	Assignee:	Raghavendra G <rgowdapp>
Status:	CLOSED CURRENTRELEASE	QA Contact:
Severity:	high	Docs Contact:
Priority:	urgent
Version:	mainline	CC:	amarts, gluster-bugs
Target Milestone:	---
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:	glusterfs-3.4.0	Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2013-07-24 17:54:35 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:
Bug Depends On:
Bug Blocks:	817967

Description Anush Shetty 2012-01-23 12:48:23 UTC

Description of problem: I was doing volume set operations in a lot while doing some I/O on the mount point. This pushed the memory usage of glusterfs fuse process to 2.5g. While the volume set operations and the I/O on the client were stopped, there was drop in the memory usage. But even after doing drop_caches, the fd count for the process was 1024.


Version-Release number of selected component (if applicable): 3.3.0qa19


How reproducible: Do volume set operations in a loop while doing some I/O on the fuse client.

Comment 1 Raghavendra G 2012-02-22 09:56:44 UTC

Anush,

Were those fds network sockets used by client process? You can find that by,

# ls -l /proc/<pid of glusterfs-client>/fd/

If so, patches,
http://review.gluster.com/#change,2734,patchset=1
http://review.gluster.com/#change,2735

are aimed at solving this problem.

Comment 2 Anush Shetty 2012-02-23 11:09:45 UTC

Hi Raghavendra,

I observed the same problem again. The fd count kept increasing while the volume set command was being run.

-
Anush

Comment 3 Anand Avati 2012-02-29 10:16:21 UTC

CHANGE: http://review.gluster.com/2831 (protocol/client: send unique key to server during handshake) merged in master by Vijay Bellur (vijay)

Comment 4 Anush Shetty 2012-04-25 12:32:11 UTC

We don't see the same kind of behaviour in 3.3.0qa38. So moving it to verified.